WO2018092258A1

WO2018092258A1 - Air conditioner and air-conditioning system

Info

Publication number: WO2018092258A1
Application number: PCT/JP2016/084231
Authority: WO
Inventors: 吉秋小泉; 雅史冨田
Original assignee: 三菱電機株式会社
Priority date: 2016-11-18
Filing date: 2016-11-18
Publication date: 2018-05-24
Also published as: CN109937331A; JP6625239B2; US11486594B2; EP3348924A1; CN109937331B; US20200049363A1; EP3348924A4; JPWO2018092258A1; EP3348924B1

Abstract

This air conditioner comprises: an outdoor unit and an indoor unit that are provided with devices and pipes which form a refrigerant circuit; and a remote controller that is connected to the indoor unit. Sensors for detecting the temperature states of the devices and pipes are provided to each of the outdoor unit and the indoor unit, and a memory for storing sensor information showing the results detected by the sensors and control information showing the controlled states of the devices is provided to the outdoor unit or the indoor unit. The remote controller has: a neural-network calculation means to calculate the possibility of failure factors by using a neural network with the information showing the state of each portion based on the sensor information and control information acquired at the same time from the memory as input values and estimated failure factors as output values; and a display means for displaying the results calculated by the neural-network calculation means.

Description

Air conditioner and air conditioning system

The present invention relates to an air conditioner and an air conditioning system for estimating a failure factor.

A remote controller provided in a conventional air conditioner management system is connected to an indoor unit via a transmission line, and holds data such as an operation history and a fault code received from the indoor unit. In addition, the remote controller can transmit the stored data to a mobile terminal such as a user's mobile phone.

Furthermore, the remote controller includes operating information such as the set temperature of the air conditioner, information indicating the type of model, operating time of the air conditioner, current consumption and rotation speed of the compressor in the outdoor unit, code indicating the cause of the failure, and It holds data such as outside temperature. Since such various information can be transmitted from the mobile terminal to a service store that performs maintenance or the like, the service store specifies service information related to inspection based on the received various information (for example, Patent Document 1). reference).

In addition, a hot water supply device has been proposed that acquires data composed of detection signals of various sensors and command values such as burner operation at preset times, and identifies a failure site based on the acquired data (for example, Patent Document 2). In this hot water supply device, data consisting of the detection signals and command values of the various sensors described above are stored as saved data, and when an abnormality occurs in the device, the latest data of the saved data before the occurrence of the abnormality of the device, The number corresponding to the device abnormality is displayed on the display unit of the remote control. As a result, workers who repair hot water supply equipment can easily identify the faulty part of the hot water supply equipment based on the displayed contents without using a dedicated measuring instrument, etc. Etc. can be achieved.

JP 2009-14233 A Japanese Patent No. 3897680

By the way, in the invention described in Patent Document 1, the obtained operation history such as the current value of the compressor is compared with the average value in the same region, the same model, and the same condition. As a result of the comparison, when the increase or decrease in the operation history with respect to the average value exceeds a preset threshold value, it is considered that there is some abnormality in the compressor or other parts.

For example, when “the air conditioner does not cool”, it is necessary to observe the operating state of the compressor or the refrigerant circuit. For this reason, it is difficult to specify the cause of failure only by performing threshold determination for individual components. In addition, since the failure factor is affected by the difference due to the installation condition and the difference depending on the model, the failure factor cannot be accurately estimated by displaying the fixed failure code.

Furthermore, in the invention described in Patent Document 2, since the operator specifies the failure part based on the information displayed on the display unit, whether or not the failure part can be specified accurately depends on the skill and experience of the operator. Resulting in. That is, in the invention described in Patent Document 2, it is difficult to accurately estimate the failure site.

The present invention has been made in view of the above problems, and an object of the present invention is to provide an air conditioner and an air conditioning system capable of accurately estimating a failure factor and improving the detection accuracy of an abnormal location. To do.

An air conditioner according to the present invention is an air conditioner including an outdoor unit and an indoor unit provided with respective devices and pipes forming a refrigerant circuit, and a remote controller connected to the indoor unit, Sensors for detecting the temperature state of the device and the pipe are provided in each of the outdoor unit and the indoor unit, and store sensor information indicating a detection result by the sensor and control information indicating a control state of the device. A memory is provided in the outdoor unit or the indoor unit, and the remote controller uses information indicating the state of each unit based on the sensor information and the control information at the same time acquired from the memory as an input value, A neural network that uses the estimated failure factor as an output value and calculates the possibility of the failure factor using a neural network. And network operation means, and has a display means for displaying a calculation result by the neural network calculating means.

As described above, according to the present invention, by estimating a failure factor using a neural network, it is possible to accurately estimate the failure factor and improve the detection accuracy of the abnormal part.

It is a block diagram which shows an example of a structure of the air conditioner which concerns on Embodiment 1. FIG. It is the schematic for demonstrating the neural network calculation process performed by the neural network calculation means of FIG. It is the schematic for demonstrating the result of a neural network calculation process. It is a block diagram which shows an example of a structure of the air conditioner which concerns on Embodiment 2. FIG. It is a block diagram which shows an example of a structure of the air conditioner which concerns on Embodiment 3. FIG.

Embodiment 1 FIG.
Hereinafter, the air conditioner according to the first embodiment will be described.

[Configuration of air conditioner]
FIG. 1 is a block diagram illustrating an example of the configuration of the air conditioner 1 according to the first embodiment.
As shown in FIG. 1, the air conditioner 1 includes an outdoor unit 10, an indoor unit 20, and a remote controller (hereinafter referred to as “remote controller” as appropriate) 30. The outdoor unit 10 and the indoor unit 20 are connected by a first communication method using a first connection line 2 that is wired or wireless. The indoor unit 20 and the remote controller 30 are connected by a second connection method 3 using a second communication method by a wired or wireless second connection line 3.

In addition, the remote controller 30 is connected to the information terminal 40 using the third communication method by the third connection line 4 by radio. As the third communication method, for example, short-range wireless communication using BLE (Bluetooth (registered trademark) Low Energy) technology can be used. In addition to the information terminal 40, the remote controller 30 can be connected to a general-purpose device such as a temperature / humidity sensor (not shown) installed in the air-conditioning target space using the third communication method.

The information terminal 40 is capable of notifying the user of information related to the air conditioner 1 such as the control state of each part in the air conditioner 1. As the information terminal 40, for example, a mobile terminal such as a smartphone, a tablet terminal, and a notebook PC (Personal Computer) can be used. The information terminal 40 is not limited to this, and a terminal that is fixedly installed such as a stationary PC may be used.

In the air conditioner 1, each device and piping forming a refrigerant circuit such as a compressor and a heat exchanger are provided in the outdoor unit 10 or the indoor unit 20. In FIG. Only the part relevant to the feature of FIG. 2 is shown, and the other parts are not shown and described. Further, in this example, the air conditioner 1 is provided with one outdoor unit 10 and one indoor unit 20, but is not limited thereto. For example, one of the outdoor unit 10 and the indoor unit 20 is provided. One side may be provided in plural, or both may be provided in plural. That is, the number of outdoor units 10 and indoor units 20 can be appropriately determined according to the situation where the air conditioner 1 is installed.

(Outdoor unit)
The outdoor unit 10 includes one or a plurality of sensors 11, a microcomputer (hereinafter appropriately referred to as “microcomputer”) 12, a first communication unit 13, a memory 14, and a compressor 15 and an expansion device that form a refrigerant circuit. A valve 16 is provided.

Sensor 11 is installed in each part of outdoor unit 10, and detects the state of a measuring object. Specifically, for example, each sensor 11 is a temperature sensor, and detects the temperature state of each part such as the outside air temperature, the compressor 15 temperature, and the piping temperature. The detected temperature information and the like are supplied to the microcomputer 12 as sensor information related to the outdoor unit 10 (hereinafter referred to as “outdoor unit sensor information” as appropriate). Note that the sensor 11 is not limited to a temperature sensor, and for example, a pressure sensor or the like may be used to detect the pressure of each part.

The microcomputer 12 controls the entire outdoor unit 10 such as controlling the operation of devices forming a refrigerant circuit such as the compressor 15 and the expansion valve 16. For example, the microcomputer 12 instructs the compressor frequency of the compressor 15 and the opening degree of the expansion valve 16.

Further, the microcomputer 12 sets and changes the state of the outdoor unit 10 based on the control instruction information received from the remote controller 30 via the indoor unit 20. Further, the microcomputer 12 acquires outdoor unit sensor information detected by the sensor 11 and control information indicating the control state of the equipment provided in the outdoor unit 10 such as the compressor frequency of the compressor 15, and the memory described later 14 and the communication of the first communication means 13 to be described later. Details of the control instruction information will be described later.

The first communication means 13 controls communication performed with the indoor unit 20 using the first communication method based on a command from the microcomputer 12. For example, the first communication unit 13 receives sensor information (hereinafter referred to as “indoor unit sensor information”) relating to the indoor unit 20 supplied from the indoor unit 20 and sends the received indoor unit sensor information to the microcomputer 12. Supply.

Further, the first communication means 13 receives control instruction information from the remote controller 30 via the indoor unit 20 and supplies the received control instruction information to the microcomputer 12. Further, the first communication means 13 receives outdoor unit sensor information, indoor unit sensor information, and control information held in a memory 14 described later from the microcomputer 12 and transmits them to the indoor unit 20. Hereinafter, “outdoor unit sensor information and indoor unit sensor information” will be collectively referred to as “sensor information” as appropriate.

The memory 14 is data holding means for holding various data. The memory 14 writes and reads outdoor unit sensor information detected by the sensor 11 under the control of the microcomputer 12. Further, the memory 14 writes and reads the indoor unit sensor information such as the suction temperature of the indoor unit 20 and the temperature of the piping acquired through the first communication unit 13 under the control of the microcomputer 12.

(Indoor unit)
The indoor unit 20 includes one or more sensors 21, a microcomputer 22, second communication means 23, third communication means 24, and a memory 25.

Sensor 21 is installed in each part of indoor unit 20, and detects the state of a measuring object. Specifically, for example, each sensor 21 is a temperature sensor, and detects the temperature state of each part such as the air suction temperature and the piping temperature in the air-conditioning target space. The detected temperature information and the like are supplied to the microcomputer 22 as indoor unit sensor information. Note that the sensor 21 is not limited to a temperature sensor, and for example, a pressure sensor or the like may be used to detect the pressure of each part.

The microcomputer 22 controls the whole indoor unit 20 such as performing operation control of devices forming the refrigerant circuit. For example, the microcomputer 22 sets and changes the state of the indoor unit 20 based on control instruction information received from the remote controller 30 described later, and transfers the received control instruction information to the outdoor unit 10 as necessary. The microcomputer 12 acquires indoor unit sensor information indicating the state of each part such as the suction temperature and the pipe temperature detected by the sensor 21, controls writing to a memory 25 described later, and second communication described later. The communication between the means 23 and the third communication means 24 is controlled.

The second communication means 23 controls communication performed with the outdoor unit 10 using the first communication method based on a command from the microcomputer 22. For example, the second communication means 23 receives the indoor unit sensor information detected by the sensor 21 and the control instruction information from the remote controller 30 from the microcomputer 22 and transmits them to the outdoor unit 10. The second communication unit 23 receives sensor information and control information from the outdoor unit 10 and supplies the received information to the microcomputer 22.

The third communication means 24 controls communication performed with the remote controller 30 using the second communication method based on a command from the microcomputer 22. For example, the third communication unit 24 receives control instruction information from the remote controller 30 and supplies the received control instruction information to the microcomputer 22. The third communication unit 24 receives sensor information and control information from the microcomputer 22 and transmits them to the remote controller 30.

The memory 25 is data holding means for holding various data. The memory 25 writes and reads the indoor unit sensor information detected by the sensor 11 under the control of the microcomputer 22.

(Remote controller)
The remote controller 30 includes a fourth communication unit 31, a microcomputer 32, a memory 33, a fifth communication unit 34, a display unit 35, and an operation unit 36.

The fourth communication unit 31 controls communication performed with the indoor unit 20 using the second communication method based on a command from the microcomputer 32. For example, the fourth communication unit 31 transmits control instruction information for controlling operations of the outdoor unit 10 and the indoor unit 20 from the microcomputer 32 to the indoor unit 20. The fourth communication means 31 receives sensor information and control information from the indoor unit 20 and supplies them to the microcomputer 32.

The microcomputer 32 controls the entire remote controller 30 based on a user operation on an operation means 36 described later. For example, the microcomputer 32 generates control instruction information for controlling the operations of the outdoor unit 10 and the indoor unit 20 based on an operation signal obtained by a user operation.

The microcomputer 32 is provided with a neural network calculation means 39. The neural network calculation means 39 probabilistically estimates the state of the air conditioner 1 using a neural network. Specifically, the neural network calculation means 39 determines whether the operation of the air conditioner 1 is normal or some trouble may occur based on various information acquired via the indoor unit 20. To do. Then, the microcomputer 32 supplies determination information indicating the determination result by the neural network calculation means 39 to the memory 33. The details of the calculation processing by the neural network calculation means 39 will be described later.

The memory 33 is data holding means for holding various data. Under the control of the microcomputer 32, the memory 33 writes and reads determination information indicating the result of determination by the neural network calculation means 39.

The fifth communication means 34 controls communication performed with the information terminal 40 using the third communication method based on a command from the microcomputer 32. For example, the fifth communication unit 34 transmits the determination information read from the memory 33 under the control of the microcomputer 32 to the information terminal 40. The information terminal 40 transmits the determination information received from the remote controller 30 to the cloud 50 connected via the network 5 such as the Internet and stores it on the cloud 50.

The display means 35 is composed of, for example, an LCD (Liquid Crystal Display), an organic EL (Electro Luminescence) display, and the like, and displays a determination result based on the determination information. As the display unit 35, not only the determination result but also a touch panel display in which a touch panel having a touch sensor is stacked on an LCD or an organic EL display can be used.

The operation means 36 is provided with various buttons or keys used for operating the air conditioner 1, and outputs an operation signal corresponding to the operation on each button or key. As described above, when the display unit 35 is a touch panel display, various buttons or keys may be displayed on the display unit 35 as software buttons or software keys.

[Neural network calculation processing]
Next, a neural network calculation process performed by the neural network calculation means 39 will be described. As described above, the neural network calculation unit 39 probabilistically estimates the state of the air conditioner 1 using the neural network. Such state determination of the air conditioner 1 is performed, for example, when an abnormality is detected while the air conditioner 1 is operating.

FIG. 2 is a schematic diagram for explaining a neural network calculation process performed by the neural network calculation means 39 of FIG. As shown in FIG. 2, the neural network 100 used in the first embodiment is a hierarchical network including an input layer 110, an intermediate layer 120, and an output layer 130 composed of a plurality of units. In this example, the intermediate layer 120 includes two layers, a first intermediate layer 121 and a second intermediate layer 122.

The input layer 110 transmits a signal based on the input information to the intermediate layer 120. Each of the units constituting the input layer 110 is coupled to all the units constituting the first intermediate layer 121 which is the next layer. The intermediate layer 120 performs arithmetic processing based on a signal input from the immediately preceding layer and outputs a calculation result. Each of the units constituting the intermediate layer 120 is combined with all the units constituting the next layer. The output layer 130 performs arithmetic processing based on the signal input from the immediately preceding second intermediate layer 122 and outputs the calculation result as an output signal.

In the neural network 100 according to the first embodiment, information indicating the state of the air conditioner 1 is input to the input layer 110 as an input signal. Specifically, for example, “compressor frequency”, “high pressure”, “low pressure”, and “superheat” at the same time are input to the input layer 110 as input signals. These input signals can be acquired based on sensor information and control information held in the memory 14 of the outdoor unit 10.

Also, from the output layer 130, a failure factor in the air conditioner 1 is output as an output signal. Specifically, for example, “normal”, “evaporator airflow reduction”, “compressor abnormality”, “refrigerant shortage”, and “condenser airflow reduction” as estimated failure factors are output signals from the output layer 130. Is output.

In the neural network 100, when signals are transmitted from each layer to the next layer, corresponding weights w _ij are accumulated for the transmitted signals. The weight w _ij is preset in the neural network calculation process, and reflects the content learned by the neural network 100. The subscript “i” in the weight w _ij indicates the unit number in the layer that is the start point of the network, and the subscript “j” indicates the unit number in the layer that is the end point of the network. This weight w _ij is stored in the memory 33 of the remote controller 30 as a weighting table. Details of the weight w _ij will be described later.

Next, the flow of the neural network calculation process using the neural network 100 will be described with reference to FIG. First, in the neural network 100 shown in FIG. 2, information indicating each of “compressor frequency”, “high pressure”, “low pressure”, and “superheat” is input as an input signal to each unit of the input layer 110. . Each unit of the input layer 110 transmits the received input signal to each unit of the first intermediate layer 121. That is, all input signals input to the input layer 110 are input to each unit of the first intermediate layer 121.

Next, each unit of the first intermediate layer 121 integrates the input signal received from each unit of the input layer 110 and the weight w _ij corresponding to the input signal, and all signals obtained by the integration are obtained. Generate the added signal. Each unit of the first intermediate layer 121 transmits the first intermediate layer signal obtained based on the generated signal to each unit of the second intermediate layer 122.

Each unit of the second intermediate layer 122 integrates the first intermediate layer signal received from each unit of the first intermediate layer 121 and the weight w _ij corresponding to the signal, and all signals obtained by the integration. To generate a signal. Each unit of the second intermediate layer 122 transmits a second intermediate layer signal obtained based on the generated signal to each unit of the output layer 130.

Each unit of the output layer 130 integrates the second intermediate layer signal received from each unit of the second intermediate layer 122 and the weight w _ij corresponding to the signal, and adds all signals obtained by the integration. Generated signal. Each unit of the output layer 130 outputs an output signal obtained based on the generated signal. At this time, the sum of the values of the output signals output from the output layer 130 is set to “1”.

As described above, the neural network calculation unit 39 acquires the output signal based on the input signal obtained based on the sensor information and the control information and the weight w _ij obtained by referring to the weighting table stored in the memory 33. To do.

FIG. 3 is a schematic diagram for explaining the result of the neural network calculation process. FIG. 3 is a graph showing a possibility of an abnormality that may occur when an abnormality is detected during operation of the air conditioner 1, for example. In this example, among the output signals from the neural network 100, values obtained by normalizing the values of the output signals corresponding to the respective failure factors with reference to the value indicating “normal” are shown. That is, the failure factor corresponding to the output signal having a value larger than the value “1” indicating “normal” may be a cause of abnormality. Therefore, in the example illustrated in FIG. 3, “shortage of refrigerant” indicates the highest possibility of being a failure factor when an abnormality is detected. A graph indicating the determination result of such a failure factor is displayed on the display means 35 of the remote controller 30, for example. Thereby, at the time of maintenance etc., an operator can estimate a failure factor easily, and maintainability can be improved.

(Weight change)
In the first embodiment, it is possible to change the weight w _ij included in the weighting table to an optimum one by answering whether or not the calculation result obtained as described above is correct. Here, “the calculation result is correct” means that the cause of the abnormality is the most likely cause of the failure factor obtained by the neural network calculation processing.

The updated weight w _ij in this case is calculated by, for example, error back propagation. Note that error back propagation is a method that is generally used when calculating weights in the neural network 100, and thus description thereof is omitted here.

The recalculation of the weight w _ij using error back propagation or the like is performed by, for example, an external PC or the like connected to the network 5. For example, when the operator operates the operation means 36 and inputs an answer to the calculation result, the remote controller 30 transmits answer information indicating the input answer to an external PC. As a result, the PC recalculates the weight w _ij based on the response information, sensor information, and control information using back propagation.

The remote controller 30 receives the recalculated weight w _ij from the external PC via the network 5, the information terminal 40, and the fifth communication means 34. Then, the microcomputer 32 of the remote controller 30 updates the weighting table by storing the received weight w _ij in the weighting table stored in the memory 33.

As described above, the air conditioner 1 according to the first embodiment includes the outdoor unit 10 and the indoor unit 20 provided with the devices and pipes forming the refrigerant circuit, and the remote controller 30 connected to the indoor unit 20. Are provided in each of the outdoor unit 10 and the indoor unit 20, and sensor information indicating detection results by the

sensors

11 and 21 and a control state of the device are provided. The outdoor unit 10 or the indoor unit 20 is provided with a

memory

14 or 25 that stores control information to be shown. The remote controller 30 uses information indicating the state of each part based on sensor information and control information at the same time acquired from the

memory

14 or 25 as an input value, and uses an estimated failure factor as an output value. It has a neural network calculation means 39 for calculating the possibility of a failure factor, and a display means 35 for displaying a calculation result by the neural network calculation means 39.

Thus, in the first embodiment, since the failure factor is probabilistically estimated using the neural network 100, it is possible to accurately estimate the failure factor and improve the detection accuracy of the abnormal part.

Moreover, since the value of the weight w _ij used in the arithmetic processing by the neural network 100 is recalculated and updated based on the response information indicating whether the failure factor is correct, sensor information, and control information, the accuracy of the failure factor estimation is increased. It can be improved further. Further, by performing recalculation of the value of the weight w _ij described above by an external PC or the like, a low-performance microcomputer 32 of the remote controller 30 can be used, and as a result, the cost can be reduced.

Embodiment 2. FIG.
Next, an air conditioner according to Embodiment 2 will be described. The air conditioner according to the second embodiment is different from the above-described first embodiment in that the information terminal 40 includes a neural network calculation unit.

FIG. 4 is a block diagram showing an example of the configuration of the air conditioner 1 according to the second embodiment. As shown in FIG. 4, in the air conditioner 1 according to the second embodiment, the information terminal 40 is provided with a neural network calculation means 49. In the following description, portions common to the above-described first embodiment are denoted by the same reference numerals and description thereof is omitted.

The microcomputer 32 in the remote controller 30 receives the sensor information and the control information transmitted from the outdoor unit 10 via the fourth communication unit 31 and supplies them to the fifth communication unit 34. The fifth communication unit 34 performs communication processing similar to that of the first embodiment, and transmits the sensor information and control information received from the microcomputer 32 to the information terminal 40.

The information terminal 40 performs a neural network calculation process based on the sensor information and control information received from the remote controller 30. The arithmetic processing performed by the neural network arithmetic means 49 is the same as the arithmetic processing by the neural network arithmetic means 39 in the first embodiment.

When the neural network calculation process is performed by the information terminal 40, the remote controller 30 receives information indicating the possibility of a failure factor as a calculation result from the information terminal 40 via the fifth communication unit 34. The microcomputer 32 causes the display unit 35 to display information indicating the possibility of the failure factor received from the information terminal 40.

At this time, the weighting table used for the neural network calculation process is held by, for example, the application of the information terminal 40, and whether or not the calculation result obtained by the calculation process is correct as in the first embodiment. By answering, the weight w _ij can be updated. Therefore, when the reply information from the remote controller 30 is transmitted to an external PC and the weight w _ij is recalculated by the PC, the information terminal 40 obtains the weight w _ij from the PC as in the first embodiment. And update the weighting table.

The updated weight w _ij is acquired by the information terminal 40 by, for example, manual input by the user to the information terminal 40, reading of a QR (Quick Response) code (QR code is a registered trademark), and USB (Universal Serial Bus) connection. Or by using an input / output interface such as a network connection.

As described above, the second embodiment can provide the same effects as the first embodiment. In addition, by performing neural network calculation processing with the information terminal 40 instead of the remote controller 30, the load on the microcomputer 32 in the remote controller 30 can be reduced.

Embodiment 3 FIG.
Next, an air conditioner according to Embodiment 3 will be described. The air conditioner according to the third embodiment is different from the first and second embodiments described above in that a neural network calculation unit is provided on the cloud 50.

FIG. 5 is a block diagram showing an example of the configuration of the air conditioner 1 according to the third embodiment. As shown in FIG. 5, in the air conditioner 1 according to the third embodiment, a neural network calculation unit 59 is provided on the cloud 50. In the following description, parts common to those in the first and second embodiments described above are denoted by the same reference numerals and description thereof is omitted.

The microcomputer 32 in the remote controller 30 receives the sensor information and the control information transmitted from the outdoor unit 10 via the fourth communication unit 31 and supplies them to the fifth communication unit 34. The fifth communication unit 34 performs the communication process in the first embodiment and transmits the sensor information and control information received from the microcomputer 32 to the information terminal 40.

The information terminal 40 transmits the sensor information and control information received from the remote controller 30 to the cloud 50 via the network 5. The cloud 50 performs a neural network calculation process based on the sensor information and control information received from the information terminal 40. The arithmetic processing performed by the neural network arithmetic means 59 is the same as the arithmetic processing by the neural network arithmetic means 39 in the first embodiment and the neural network arithmetic means 49 in the second embodiment.

When the neural network calculation process is performed by the cloud 50, the remote controller 30 receives information indicating the possibility of a failure factor as a calculation result from the cloud 50 via the information terminal 40 and the fifth communication unit 34. The microcomputer 32 causes the display unit 35 to display information indicating the possibility of the failure factor received from the cloud 50.

At this time, the weighting table used for the neural network calculation process is held by, for example, the application of the cloud 50, and whether or not the calculation result obtained by the calculation process is a correct answer as in the first and second embodiments. , The weight w _ij can be updated. Accordingly, when the reply information from the remote controller 30 is transmitted to an external PC and the weight w _ij is recalculated by the PC, the cloud 50 receives the weight w _ij from the PC as in the first and second embodiments. Acquire and update the weighting table.

As described above, the flow path switching valve according to the third embodiment can achieve the same effects as those of the first embodiment. In addition, by performing neural network calculation processing with the cloud 50 instead of the remote controller 30, the load on the microcomputer 32 in the remote controller 30 can be reduced.

While the first to third embodiments have been described above, the present invention is not limited to the first to third embodiments described above, and various modifications and applications can be made without departing from the scope of the present invention. . For example, in the above-described example, both the outdoor unit sensor information and the indoor unit sensor information have been described as being stored in the memory 14 of the outdoor unit 10. However, the present invention is not limited thereto, and both the outdoor unit sensor information and the indoor unit sensor information are stored. May be stored in the memory 25 of the indoor unit 20.

In the above-described example, the case where the number of intermediate layers 120 is two has been described. However, the number of intermediate layers 120 is not limited to this. For example, the number of intermediate layers 120 may be one or more. The number of intermediate layers 120 can be appropriately set in consideration of the accuracy of the state determination of the air conditioner 1 and the like.

Further, the input signal in the neural network calculation process is not limited to the above-described example, and for example, the discharge temperature of the compressor 15, the evaporation temperature of the heat exchanger, the outside air temperature, the set temperature, the opening degree of the expansion valve 16, or the like. A signal indicating this information may be used as an input signal. Further, the input signal to the input layer 110 may be based on a plurality of pieces of information acquired at preset time intervals such as 1 minute intervals.

Furthermore, in the above-described example, the failure factor is estimated when an abnormality of the air conditioner 1 is detected. However, the present invention is not limited to this. For example, at the time of installation and maintenance of the air conditioner 1, Similarly, the failure factor can be estimated.

1 air conditioner, 2nd connection line, 3rd connection line, 4th 3rd connection line, 5 network, 10 outdoor unit, 11 sensor, 12 microcomputer, 13 1st communication means, 14 memory, 15 compressor, 16 expansion valve, 20 indoor unit, 21 sensor, 22 microcomputer, 23 second communication means, 24 third communication means, 25 memory, 30 remote controller, 31 fourth communication means 31, 32 micro Computer, 33 memory, 34 fifth communication means, 35 display means, 36 operation means, 39, 49, 59 neural network operation means, 40 information terminal, 50 cloud, 100 neural network, 110 input layer, 120 intermediate layer, 121 1st intermediate layer, 122 2nd intermediate layer, 130 The output layer.

Claims

An air conditioner including an outdoor unit and an indoor unit provided with each device and piping forming a refrigerant circuit, and a remote controller connected to the indoor unit,
Sensors for detecting the temperature state of the equipment and the piping are provided in each of the outdoor unit and the indoor unit,
A memory for storing sensor information indicating a detection result by the sensor and control information indicating a control state of the device is provided in the outdoor unit or the indoor unit,
The remote controller is
The information indicating the state of each unit based on the sensor information and the control information at the same time acquired from the memory is used as an input value, and an estimated failure factor is used as an output value. Neural network computing means for computing sex;
An air conditioner having display means for displaying a calculation result by the neural network calculation means.
The neural network is
An input layer into which the input value is input;
An intermediate layer coupled to the input layer;
An output layer coupled with the intermediate layer,
The weight value when the layers are joined is calculated externally,
The neural network calculation means includes
The air conditioner according to claim 1, wherein the calculated weight value is acquired from the outside and updated.
The air conditioner according to claim 2, wherein the weight value is calculated based on information indicating whether the failure factor is correct, the sensor information, and the control information.
The display means includes
The air conditioner according to any one of claims 1 to 3, wherein a graph indicating the possibility of the failure factor is displayed on the basis of a normal state.
The air conditioner according to any one of claims 1 to 4,
An information terminal for notifying information on the air conditioner,
The neural network calculation means is an air conditioning system provided in the information terminal instead of the remote controller.
The information terminal is connected to the cloud via a network,
6. The air conditioning system according to claim 5, wherein the neural network calculation means is provided in the cloud instead of the information terminal.